This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

Research Article

www.acsami.org

3D-Printed Self-Folding Electronics

Subramanian Sundaram,*,, David S. Kim, Marc A. Baldo, Ryan C. Hayward,
and Wojciech Matusik,

Computer Science and Articial Intelligence Laboratory and Electrical Engineering and Computer Science Department,
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States

Polymer Science and Engineering Department, University of Massachusetts, Amherst, Massachusetts 01003, United States
*
S Supporting Information

ABSTRACT: Self-transforming structures are gaining prom-

inence due to their general ability to adopt programmed
shapes each tailored for specic functions. Composites that
self-fold have so far relied on using the stimuliresponsive
mechanisms focusing on reversible shape change. Integrating
additional functions within these composites can rapidly
enhance their practical applicability; however, this remains a
challenging problem. Here, we demonstrate a method for
spontaneous folding of three-dimensional (3D)-printed
composites with embedded electronics at room temperature.
The composite is printed using a multimaterial 3D-printing
process with no external processing steps. Upon peeling from the print platform, the composite self-shapes itself using the
residual forces resulting from polymer swelling during the layer-by-layer fabrication process. As a specic example, electrochromic
elements are printed within the composite and can be electrically controlled through its folded legs. Our shape-transformation
scheme provides a route to transform planar electronics into nonplanar geometries containing the overhangs. Integrating
electronics within complex 3D shapes can enable new applications in sensing and robotics.
KEYWORDS: self-folding, multimaterial 3D-printing, exible electronics, electrochromic pixels, robotics

INTRODUCTION
Programmable matter and self-assembled structures have
highlighted a range of applications that unify physical shape
and overall function.1,2 Emerging advances in robotics,3
photovoltaics,4 antennas,5 biomimetic imagers,6 and inertial
measurement units7 present a compelling class of devices where
electronics have evolved to nontraditional complex architec-
tures. Adapting contemporary planar electronics that have high
performance to three-dimensional (3D) geometries has proved
to be a challenging problem. This is currently being tackled
using a variety of techniques, such as using droplet wetting to
induce self-folding,4 manual folding of silicon pieces with
interlocks,7 and compressive buckling of thin silicon mem-
branes bonded on prestretched materials.8 The general
approach thus far is to create the electrical device on a planar
silicon wafer using microfabrication techniques and subse-
quently shape it using dierent methods. These shaping
techniques have the advantage of being directly applicable to
planar-fabricated silicon-based electronics. However, they tend
to be device specic, constrained in possible geometries, or not
easily scalable. A process that creates an electrical function
within a complex 3D structure seamlessly is of signicant
practical importance, and is therefore an active area of research.
Outside the connes of traditional electronic device
fabrication, a broad set of origami- and kirigami-inspired
techniques have been extensively explored as the potential
routes to fabricate complex 3D structures thus far.2,3,915
Demonstrations of material folding have spanned several orders
of magnitude in length scales, utilizing a disparate set of
mechanisms including9 controlled buckling,8 shape-memory
eect in composites,10,11 pneumatically driven transforma-
tions,12 thermally controlled swelling in hydrogels,13 photo-
induced folding and manipulation,1418 residual stress-induced
curling,19 and controlled formation of creases.20 Recently,
additive manufacturing has also been used to create 3D
structures using shape transformation (four-dimensional (4D)-
printing).2123 This is in part driven by the high-speed
capabilities of 3D-printing24 and the ability to assemble a set
of diverse materials simultaneously.2528 Folding at, printed
designs into complex structures has additional benets over
directly printing the nal 3D structure, typically in the overall
fabrication speed, cost, and ease of scaling. Furthermore, when
solvent-based inks are used (as is typical for a majority of
electrically important materials), folded geometries may not be
directly 3D-printed due to challenges from droplet spreading
and nonuniform lm thicknesses. This has generated
considerable recent interest in 4D-printing; structures that
change dynamically allows the shape to be controlled over
Received: July 17, 2017
Accepted: August 21, 2017
Published: August 21, 2017

XXXX American Chemical Society A DOI: 10.1021/acsami.7b10443

ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces Research Article

time.2934 Although there have been attempts to embed

packaged electronics inside 3D-printed structures35,36 to
improve their overall functionality,37 achieving fully 3D-printed
electronics that self-fold is a current challenge. As a rst step in
this direction, Zarek et al. have recently inkjet-printed
conductors on top of attened 3D-printed shapes that can
then be restored to their original shapes using the shape-
memory eect.38 Similarly, electrically conductive shape-
memory materials have also been 3D-printed via direct ink
writing.39 Despite the tremendous interest in shape trans-
formation in 3D-printed structures, there are several practical
challenges that exist in expanding these structures to include
printed electronic devices. The two common classes of
mechanisms used for self-folding in 3D-printed structures are
based on the temperature-controlled shape-memory eect in
polymers and the controllable swelling of hydrogels in water
and other liquids.40 The main focus of these approaches is the
ability to achieve reversible self-folding, which is a signicant
advantage. The required stimuli, that is, temperature cycling
and immersion in liquids, however, may not desirable for use
with electronic devices. This is particularly relevant if the
electrically active materials of the self-shaping composite may
degrade with temperature cycling or exposure to humidity.
Controllable folding with the ability to engineer stress at
multiple positions is more favorable for electronics applications
that do not require reversible shape change. Recently, Ding et
al. reported a folding method using a temperature input in a
system of polymers with dierent coecients of thermal
expansion to accelerate folding based on an inbuilt strain.41
Figure 1. Schematic representation of spontaneously folding
This is an elegant method that can create rapid folds. However,
electronics achieved by multimaterial 3D-printing. The composites
the use of a commercial printer with proprietary inks (that by containing active electronics, metallic traces, and the substrate are fully
default uses a attening roller) makes it challenging to identify 3D-printed. On removing the 3D-printed structure from the print
the stress creation mechanism. Furthermore, electrically platform, the composite spontaneously folds in a shape based on the
relevant materials cannot be printed with this system, which locally engineered residual stress, requiring no stimulus at room
is the main goal of our work. temperature. At the heart of this scheme is a UV-curable ink

RESULTS AND DISCUSSION

Here, we present a new approach of fabricating electronic
composites that can self-fold at room temperature. We achieve
this by fully 3D-printing the electronic device and the
surrounding polymer matrix with locally engineered stress in
a single system with no external intervention. A drop-on-
demand multimaterial 3D-printing process with a voxel-level
control is used to fabricate the structures in this work.42,43
Upon removing the printed parts from the platform, they self-
fold into the preprogrammed 3D structure requiring no other
controlled stimulus. Both convex and concave folds are
achieved by directly writing the residual stress in select regions
of the printed object, as shown in the scheme in Figure 1. To
create the residual expanding tendency, we formulate an ink
containing acrylate monomers and oligomers that can be cross-
linked by UV light (detailed formulation is in Experimental
Section). The UV-cured (cross-linked) polymer swells
signicantly in the starting ink containing small molecules
(with pronounced swelling in isooctyl acrylate). Immersion in a
monomer solution has been recently used as an external
stimulus to expand a low cross-link density polymer.44 During
the layer-by-layer printing process here, each underlying cured
layer (17 m in thickness), is automatically exposed to new
ink, which diuses into the partly cross-linked layer below
before the UV light passes over it. This causes the material to
have an inherent residual stress that forces it to expand when
unconstrained (removed from the print platform). This is
B DOI: 10.1021/acsami.7b10443
containing high-elongation acrylate oligomers and short-chain
monomers (such as isooctyl acrylate). As a part of the layer-by-layer
inkjet-based 3D-printing process, each previously cured (lightly cross-
linked) layer, 17 m thick, is exposed to uncured ink from the
subsequent layers. The small molecules from the ink swell the
underlying cured layers and create a local residual stress. Both the
circular close-ups shown in the bottom only show the residual stress
material before and after curing; the rigid material is not linked to the
stress-generation mechanism. The role of each ink component is
shown in Figure S1. This material, which is used to directly write
residual stress in a volume, is combined with rigid UV-curable acrylates
to create both concave and convex folds of the desired angles by
controlling the material placement and geometry in the design. The
active electronics are linked by a stretchable conductor all printed
within the same system.
illustrated in the bottom right of the scheme in Figure 1, where
the ink before curing is shown in yellow and the lightly cross-
linked upper layers are shown in green (the colors correspond
to the same residual stress material before and after curing).
At the heart of this self-folding scheme is the swelling of the
cross-linked polymer in its starting ink components, and the
ability to deposit multiple materials. To verify that the layer-by-
layer process that facilitates swelling is indeed responsible for
the introduction of the residual stress, the ink is poured in a
mold and cured in a single step using a high-intensity (14 W
cm2 at 365 nm, UV fusion) broadband UV source. Upon
release from the mold, the cured piece does not expand but
instead shrinks slightly as a result of the formation of cross-
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces Research Article

Figure 2. Residual stress polymer and process characterization. (A) The double-bond conversion percentage is calculated as a function of the UV
light intensity and the number of curing passes from the Fourier transform infrared (FTIR) attenuated total reection (ATR) measurements on the
surface of the samples (three samples for each data point). The absorbance values corresponding to the reactive double bond CHCH2 group
are normalized with respect to the COO group to obtain the reaction conversion (40). The sample FTIR spectra are shown in Figure S2. (B)
Plot of the cure depth measured as a function UV light intensity and the number of passes. (C) Elastic modulus values are measured for the samples
after one UV curing pass at various light intensities (0.3, 0.6, and 0.9 W cm2 using the printer and 14 W cm2 using a high-intensity broadband
source, UV fusion). (D) Measured linear expansion in one dimension as a function of double-bond conversion for dierent swelling times (samples
are 125 m thick). The samples are prepared using single curing passes with varying UV light-emitting diode (LED) intensities. As printed, the
samples shrink by <1% typically. These samples with one-shot curing are expected to show isotropic swelling. The plot also shows the measured gel
fraction of these samples.

links. To quantify the swelling tendency of the cross-linked
polymer in the ink components, the samples (cured in a single
step) are allowed to swell by immersion in each of the
individual ink components. The swelling results (Figure S1)
show that SR440 (isooctyl acrylate) plays a dominant role
(causing 10% expansion in each dimension of a 1 mm thick
slab, after 1 h), whereas all of the other individual components
lead to negligible swelling over the same time. This is in part
due to the nature of the side chains of the acrylate components.
SR440, SR504, and SR313B include an isooctyl group (C8H17
chain), ethoxylated nonyl-phenol group (C23H39O4 group,
which includes a phenol group), and a dodecyl group (C12H25
chain; methacrylate), respectively. It is expected that SR440
diuses the most due to its low molecular weight and shortest
side chain, although multiple other factors may be relevant. The
expected diusion times are explained further in Figure S1. It is
C DOI: 10.1021/acsami.7b10443
important to note that during the actual printing process, the
uppermost printed layers are exposed to one pass of the 0.9 W
cm2 UV source before exposure to the fresh ink. Therefore,
the diusion is expected to be further enhanced in the printing
process compared with the single-step fully cross-linked
samples. After small molecules diuse into the underlying
layers, they are cross-linked into the existing network with the
subsequent UV-curing passes.
To gain further insights into the state of the residual stress
polymer, we measured the Fourier transform infrared (FTIR)
attenuated total reection (ATR) spectra for the samples cured
with various UV intensities and number of curing passes in the
printer. The reactive double conversion is measured by
normalizing the absorption peak of the CHCH2 group
with respect to the COO group40 (the sample FTIR
spectra are shown in Figure S2 and described in Experimental
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces Research Article

Figure 3. Temporal characteristics of the bilayer system used for creating folds. (A) Inset shows a bilayer stack where the material with the residual
stress (green, called residual stress material in text) causes the underlying substrate made of the rigid material to bend. The plot shows the time
evolution of d, the bending angle at standard room temperature (25 C) once the sample is removed from the print platform. The dierent curves
show the bending angle as a function of time for varying thickness of the material with residual stress, ts. For these tests, the thickness of the rigid
polymer is tr = 99.9 4.3 m, width W = 3.25 mm, and length Lr = 15 mm and Ls = 4 mm. The folding angle does not settle even after 3000 min.
(B) The plot shows the inverse of the creep compliance of the rigid material at 25, 35, and 45 C. The material shows a rapid increase in the
compliance until 30 min, after which the increase in creep compliance settles to a slower rate. The modulus drops by more than 1 order of
magnitude between the start and 10 h into the creep test. With the rigid material having a glass transition temperature (Tg) close to the room
temperature, the inherent stiness of the material drops drastically as the temperature increases above 25 C. The shaded region shows the time it
takes for the inverse creep compliance to drop below 100 MPa. The inset shows the relaxation modulus of the stress layer material. The relaxation
modulus shows a negligible drop after 200 min (of the stress-relaxation experiment). Further, there is no signicant change in the base stiness of the
material as the temperature is increased to 35 C because the material is well into the rubbery regime. The glass transition measurements for the two
materials are included in Figure S3.

Section). From Figure 2A, it is observed that the double-bond
conversion of the lightly cross-linked layer (one pass with the
default 0.9 W cm2 intensity source) is 77%; the samples
reach near full double-bond conversion after three passes (from
the three samples for each data point). For the measurements
in Figure 2A, the light intensity was varied below the default
intensity to understand the eect of reduced UV-curing
intensity. Here, new material was not added in between the
curing passes purely to understand the eect of multiple curing
passes. During the printing process, there is a gradient in the
received UV dosage going from the uppermost printed surface
(latest) to the rst printed layers. To perform a controlled
study of the equivalent initial strain in the samples without the
dosage gradients along the thickness, we performed subsequent
measurements of gel fraction and linear strain using single-step
cured samples of a controlled thickness. For the thickness of
layers to be meaningful, that is, without signicant gradients in
the cross-link density along the depth, we try to constrain this
value below the cure depth at each UV intensity, as shown in
Figure 2B. A typical cure depth after one pass of the UV light at
0.9 W cm2 intensity is 150 m. Figure 2C shows the
measured elastic modulus of the samples after one curing pass.
It can be observed that the modulus increases with the cross-
link density. To understand the strain in the printed layers, we
UV cured multiple samples with one pass of the UV light with
dierent intensities (controlling the double-bond conversion).
The measured increase in one dimension (averaged from four
individual samples) is plotted as a function of the double-bond
D DOI: 10.1021/acsami.7b10443
conversion as shown in Figure 2D. In these experiments, the
cured samples were soaked in the starting ink of the residual
stress material for dierent amounts of time to control the
swelling. It is observed that the resulting strain increases with
the swelling time. It is to be noted that these experiments were
performed with a spacer of thickness 125 m for handling
despite this being close to the cure depth at lower UV
intensities; the samples with 0.3 W cm2 curing were very
fragile and easy to tear. It is worth noting that the diusion time
is proportional to the square of the thickness; a 10 min swelling
experiment conducted on a 125 m sample corresponds to 10
s of swelling a 17 m layer. Additionally, the experiments are
conducted at room temperature; however, in the actual printing
process, the droplets leaving the printhead leave at an elevated
temperature and would experience an increase in diusion. The
measured increase in the length can be translated to the
isotropic increases in all of the three dimensions for single-pass
cured samples swelling in the starting ink. During the printing
process, the scenario is dierent with isotropic stresses only in
the in-plane directions but not in the out-of-plane directions.
The residual stress material is printed along with a thin, rigid
acrylate polymer material (thermoset) in a bilayer conguration
to generate folds, as shown in the inset of Figure 3A. The
expansion in the stress layer exerts a surface load on the rigid
polymer, causing the structure to bend, with a uniform radius of
curvature, rd. To characterize the folding, multiple bilayer
structures are 3D-printed and the folding angle is measured as a
function of the thickness of the stress layer over time. The
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces Research Article

Figure 4. Bilayer characteristics at elevated temperatures and design guidelines. (A) Plots show the time dependence of the folding angle of the
bilayer system used for creating folds at 35 and 45 C. The folding angle shows little variation past 100 min at 35 C. Samples approach a steady-
state folding angle after 5 min at 45 C. (B) The folding angle is shown as a function of the stress layer thickness at various time instants at 25 C.
The black curve shows the saturated folding angle at 35 C (after 300 min). Note that the optimum thickness of the stress layer for creating larger
folds changes with time (due to a larger creep compliance). (C) Finite-element analysis (FEA) predictions of the folding angle based on the
geometry and time (or equivalently elastic modulus, from Figure 3B). Comparison with the experimental values and Timoshenkos bilayer model is
explained in the Supporting Information and in Figure S5.

thickness of the rigid layer, tr, is 99.9 4.3 m, the width, W, is
4 mm, and the stress layer thickness varies from 71.95 to 270.9
m. The printed samples are peeled from their at state and
kept at a controlled temperature (25 C). The folding angle, d,
is measured as a function of time for multiple designs and
shown in Figure 3A, with each point showing measurements
from three samples. It is observed that the folding angle
increases with the stress layer thickness (in this range) until
1000 min, and the angle continues to increase over time,
showing no saturation up to 3000 min. When no stress layer is
added (as negative control samples), the rigid layers themselves
do not show any noticeable folding over time. To design new
structures, understanding this temporal behavior in the folding
angle and its relationship with geometry is essential. As a rst
step to understanding the material behavior at room temper-
ature (25 C) and establishing its relationship with the glass
transition temperature, dierential scanning calorimetry (DSC)
and dynamic mechanical analysis (DMA) are performed. The
glass transition temperatures (Tg) of the rigid material and the
E DOI: 10.1021/acsami.7b10443
stress layer are identied to be around 50 and 10 C,
respectively, with the rigid material exhibiting a broader glass
transition (measurements are in Figure S3). To understand the
relaxation time scales in the expanding residual stress polymer,
the stress-relaxation measurements are performed using DMA
and shown in the inset of Figure 3B. It is observed that the
relaxation modulus is 0.9 MPa and shows little change with
time and temperature, conrming that the material with the
residual stress is in a completely rubbery state. The time
evolution of the angle is thus more likely attributed to the rigid
polymer with its broad glass transition regime and the
proximity of Tg to room temperature. To understand the
time scale of polymer chain relaxation in the rigid material,
creep compliance measurements are performed using DMA,
and the results are shown in Figure 3B for 25, 35, and 45 C.
The inverse creep compliance drops by approximately an order
of magnitude by 600 min, and the rate of decrease rises as the
temperature is brought closer to 50 C. It is also observed that
the drop in the inverse creep compliance slows down beyond
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces Research Article

Figure 5. Self-folding composites. (A) Image shows an electrochromic element integrated in a spontaneously folding fully printed composite. The
composite consists of six materials that are printed simultaneously using no external processing (see Figure S7). The transmission through the
electrochromic ink (PEDOT:PSS) can be modulated by applying a voltage as shown in the bottom. Video S1 shows the electrochromic pixel. The
stretchable printed conductor provides electrical contacts through the folded legs. Scale bar is 10 mm. (B) Finite-element simulations using the
equivalent input strain (0.22) obtained earlier shows a close match to the deformed structure 5 min after the removal from the print substrate, as
shown in (A). The von Mises stress in the structure is seen to be concentrated in the bilayer regions of the legs. (C) Series of photographs shows the
change in the folding angles over the course of a week. (D) Printed bilayer structures with folding stops, before removal from the platform (top), and
a set of structures after reaching steady-state folding angles. Scale bar is 10 mm. The measured angles are plotted as a function of designed angles in
Figure S8.

2030 min after the application of the load. To conrm that
the change in the folding angle with time is a result of the
increasing rigid material (substrate) compliance, we repeat the
measurements of the folding angle for multiple designs at
elevated temperatures (35 and 45 C). At these slightly
elevated temperatures, the sharper drop in the inverse creep
compliance results in larger folds and saturation of the folding
angle after 100 and 10 min at 35 and 45 C, respectively, for
the same set of designs (shown in Figure 4A). It can be seen
that the eective modulus of the rigid material falls below 100
MPa at these time intervals from Figure 3B; a rst-order
explanation for this threshold is included in the Supporting
Information using energy methods. Eectively, as the
equivalent elastic modulus of the rigid material (substrate)
starts to drop over time, an increasing portion of the residual
energy stored in the stress layer (expanding material) is used in
deforming (bending) itself. Beyond this threshold modulus of
the rigid material, the energy spent in bending the rigid
substrate is negligible, and further reduction in its elastic
modulus does not result in an increase in the folding angle. It is
also evident that at elevated temperatures, increasing the
thickness of the stress layer does not necessarily lead to an
increase in the folding angles (experimental results in Figure
4A,B; simulation results in Figure S5B and explained in the
Supporting Information).
The eective residual energy stored in the stress layer can be
obtained from the folding angle measurements. Moreover, the
F DOI: 10.1021/acsami.7b10443
time evolution of the folding angle can be accounted for, purely
based on the change in the eective elastic modulus of the rigid
material (shown in Figure 3B). Therefore, the folding angle can
be mapped as a function of thickness of the stress layer as
shown in Figure 4B, where each time instant is mapped to a
particular value of the elastic modulus of the rigid material.
Given the device geometry and the material properties, the
bending angle can be obtained using Timoshenkos model of a
bilayer with two linear elastic materials45,13 as a function of an
equivalent initial strain in the expanding material (see the
Supporting Information). Here, using nite-element analysis
(FEA) on the measured geometry of the printed bilayer
devices, we estimated the equivalent isotropic initial strain in
the material to be 0.22; the simulation procedure is
elaborated and the results are compared with the exper-
imentally measured values and Timoshenkos equation in
Figure S5. FEA is used to obtain predictions of the folding
angle based on the geometry and time as shown in Figure 4C;
the correlation between the eective elastic modulus of the
rigid material and time is obtained from Figure 3B. It is
expected that the folding behavior is also sensitive to the
amount of cross-linking (based on UV curing). The UV
intensity of the LED array in the printer at the printing height is
0.9 W cm2; the eects of excessive UV dosage and high-
temperature exposure are discussed in the Supporting
Information and shown in Figure S6.
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces
Entire composites are fabricated with a multimaterial printer
(six materials in total) to demonstrate the potential for using
this material system in self-folding electronic composites.
Although active electronics have been printed previously,42 it
is important to recognize that when solvent-based inks are
printed, the underlying substrate has to be locally at to restrict
movement of the ink during the printing process. It is therefore
challenging to directly print fully folded composites with
embedded electronics containing curved structures with
overhangs. The stress layer and the rigid material are used as
the overall mechanical structure and act as a substrate for the
other materials: an elastic layer to control the surface energy,
reactive silver ink 46 to make the contacts, poly(3,4-
ethylenedioxythiophene):poly(styrene sulfonate) (PE-
DOT:PSS), and an electrolyte. The stress layer, the rigid
material, and the elastic material are UV cured. The convective
heater is used with the reactive silver ink to initiate precipitation
of the silver nanoparticles and to evaporate the solvent in the
PEDOT:PSS layers. In the nal device, the applied voltage
controls the movement of metal ions from the electrolyte into
the PEDOT:PSS channel. This de-dopes the PEDOT:PSS,
leading to a change in the optical contrast and electrical
conductivity.47 The design of the composite and the cross-
sectional schematic are described in the Supporting Informa-
tion and Figure S7. Figure 5A shows a printed self-folding
structure with an embedded electrochromic pixel using the
aforementioned inks, 5 min after removal from the print
platform; Figure S7B shows a printed device before and a week
after being removed from the platform. The nite-element
simulation of the folded structure (without any tting
parameters) shows a good match, with most of the stress
concentrated in the rigid material at the bilayer regions. The
elastic modulus of the rigid material 5 min after the application
of the load (1279 MPa, from Figure 3B) and = 0.22 are
used for the results shown in Figure 5B. The electrochromic
structure utilizes the optical contrast between the oxidized and
reduced states of the PEDOT:PSS lm that can be electrically
controlled. The stretchable conductor is subjected to
compressive and tensile strains in the adjacent folds in two
legs that provide electrical contacts. The precipitated silver
nanoparticle layers that make up the stretchable conductor
remain conductive until the mechanical breakage of the
underlying polymer (up to 40% strain).42 When the polarity
of the control voltage (1 V) is switched, the transmission
through the pixel changes, as shown in the inset at the bottom
in Figure 5A; a video recorded a day after the device was
removed from the substrate is shown in Video S1.
The shape of the composite changes over time, and Figure
5C shows the shape change in an array of prints over the course
of 1 week. Although the shape change over time may be
desirable in certain applications, it is equally desirable to
achieve xed folds at controllable angles irrespective of the
external temperature uctuations. Mechanical stops can be
designed on either side of the bilayer stack for this purpose, as
shown in Figure 5D (top). On removing the structures from
the substrate, and even when exposed to a hot air gun set to
100 C for 1 min, the folds stop at predesigned angles, as
shown in the bottom. The desired stop angles can be designed
and controlled accurately based on the geometry (Figure S8).
At rst, it appears that only small folding angles are achievable
at room temperature (25 C), and that the folding angle
continues to change for at least over 2 days (Figure 3A). It is
important to note that only rd, the radius of curvature, is
G DOI: 10.1021/acsami.7b10443
Research Article
constrained, that is, larger folding angles can be obtained by
scaling Ls. Mechanical stops can be designed to limit folding at
the desired radius of curvature (time). If the materials in the
composite do not degrade with a small temperature increase, a
small radius of curvature or accelerated folding can be achieved
(at 35 C). While generating folds from the planar composites,
it is expected that the achievable geometries are analogous to
the shapes from origami. The geometric limitations here come
from the achievable radius of curvature and the temporal
characteristics elaborated earlier. Flattening a fold typically
takes about 0.1 N of force for the structures in our work;
Figure S9 includes the experiments on unfolding forces and
Video S2 shows the related videos. It is worth noting that all of
the samples in this work rely on the residual stress created in
the in-plane printing directions for folding. During the printing
process, the uppermost printed layers are unconstrained on the
upper side and free to swell in the z direction. To control the
magnitude of the equivalent inbuilt strain, several approaches
exist. The amount of swelling can be increased by increasing
the layer printing time, leading to a larger residual stress.
Additionally, a multimaterial process like ours allows the use of
multiple inks with widely dierent swelling ratios, allowing
excellent spatial resolution in prescribing residual stress.
SUMMARY
In summary, the residual stress-based folding method described
here provides a new route to design 3D shapes folded from
planar polymer composites at room temperature. Furthermore,
the ability to control the degree of folding based on the design
geometry enables designs that are tolerant to small temperature
changes. The formulated inks and the corresponding material
characterization results here presents a complete toolkit
required to design, simulate, and fabricate new geometries.
The demonstrated self-folding electronic composite, fabricated
without using any external processing, is expected to inspire
electronics in new form factors. The addition of actuation
mechanisms can further improve the utility of these composites
in applications in robotics and humanmachine interfaces.
EXPERIMENTAL SECTION
Three-dimensional-Printing Process. The custom built drop-
on-demand multimaterial 3D printer and its typical use are as
described in a paper;42 details of the system architecture are in a
report.43 The specic details relevant in this work are summarized
here. The inkjet printheads are mounted on a carriage with the
material feeding system, UV-curing LEDs (wavelength = 365 nm,
intensity 0.9 W cm2), and forced convection heater (ceramic
heating element is maintained at 325350 C). The feeding system
heats the UV-curable inks to reduce the viscosity to the printable range
and connects them to a heated printhead (70 C). The solvent-based
inks are not heated and are directly connected to the printhead at
room temperature. The model to be printed is designed as an
STereoLithography le (STL) and a custom voxelizer is used to
breakdown the geometry to a 3D arrangement of voxels. Each layer of
the print le is completed in several inkjet printing passes (each pass
takes 15 s). After each printing pass, the newly printed droplets are
automatically UV cured in the case of UV-curable polymer inks, or
subjected to heating passes in the case of solvent inks. The typical in-
plane resolution of the printing system is 35 m; the dimensional
properties and the resolution are described in detail elsewhere.42
Multiple materials can be printed simultaneously in each pass of the
printer. All of the printing parameters, such as the pressure in the
feeding system and the number of heating/UV-curing passes, are
controlled from a printer conguration le and the printer user
interface.
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces
Ink Preparation. All of the inks used in this work are prepared by
mixing commercially available components. The material with the
residual stress is prepared by mixing 53.1 wt % CN3105 (low-viscosity
acrylic oligomer; Sartomer), 18.8 wt % SR440 (isooctyl acrylate;
Sartomer), 17.7 wt % SR504 (ethoxylated nonyl-phenol acrylate;
Sartomer), 8.85 wt % genomer 4215 (aliphatic urethane acrylate; Rahn
USA Corp.), 1 wt % Irgacure 819 (bis(2,4,6-trimethylbenzoyl)-
phenylphosphineoxide; BASF, Germany), 0.5 wt % 2-isopropylthiox-
anthone (ITX; Rahn USA Corp.), and 0.05 wt % methoxyphenol
(MEHQ; Sigma-Aldrich). The dierent components are homogenized
and left to mix overnight using a magnetic stirrer. Finally, the ink is
ltered using a 1 m lter and degassed using a vacuum pump to
remove small bubbles. A rigid material ink is obtained by mixing the
following components in a similar fashion: 57.8 wt % genomer 1117
(cyclic trimethylolpropane formal acrylate; Rahn USA Corp.), 31.3 wt
% genomer 2252/GP25 (bisphenol A epoxy diacrylate in 25%
propoxylated glycerol triacrylate; Rahn USA Corp.), 8.8 wt % Miramer
M300 (acrylic acid ester; Rahn USA Corp.), 1 wt % Irgacure 819, 1 wt
% ITX, and 0.1 wt % MEHQ. The other inks used in the electronic
compositeelastic material used to engineer the surface energy,
reactive silver ink,46 PEDOT:PSS ink, and electrolyte inkare
prepared using the previously reported procedure.42 Typically, all of
the inks are optimized for suitable rheological properties (viscosity <
15 mPa s) and used in a drop-in-ight analysis system (jetXpert;
ImageXpert Inc., Nashua, NH) to further optimize the waveform for a
satellite-free droplet generation.
Material Characterization and Folding Measurements.
Mechanical stressstrain tests for the material with residual stress
(Figure S4) and controlled deformation in the samples for Poissons
ratio estimation were performed using Instron 5944 (Instron,
Norwood, MA), a single-column table top mechanical testing system.
Stress-relaxation measurements for the residual stress material (in the
linear regime, at 2.5% strain), the creep compliance measurements for
the rigid material (at 0.5 MPa stress), and the storage and loss
modulus measurements (1 Hz cycling) for all of the materials were
performed using DMA Q800 (TA Instruments, New Castle, DE).
Dierential scanning calorimetry measurements were performed using
DSC Q100 (TA Instruments, New Castle, DE). Thickness measure-
ments for the printed samples were performed using DektakXT stylus
prolometer (Bruker Corp., Billerica, MA) and optical microscope
SZ61 (Olympus Corp., Tokyo, Japan) tted with SC30 3.3 MP
microscope digital camera (Olympus Corp., Tokyo, Japan). The
folding angles were measured using the Stream Start software
(Olympus Corp., Tokyo, Japan) from the images taken using a
Canon EOS 60D digital single-lens reex (DSLR) camera and the
SC30 microscope digital camera.
UV Intensity and Controlled Temperature Measurements.
UV light intensities were measured using a Loctite UV A/B
Radiometer Dosimeter (1390323; Loctite-Henkel, Dusseldorf, Ger-
many). The broadband high-intensity UV source used was a F300
Heraeus system with the LC6B Benchtop conveyor (Fusion UV/
Heraeus, Maryland). For the measurements of folding angles at
controlled temperatures, a universal oven UF 30 (Memmert,
Schwabach, Germany) is used to set up a controlled temperature
atmosphere. To generate the hot air temperature variation for the
samples with mechanical stops in Figure 5D, the hot air gun in X-
tronic 5000 series rework station (Model #5040-XTS; X-tronic Int.,
Lincoln, NE) is used.
FTIR Measurements and Double-Bond Conversion. FTIR
spectra were obtained using the Thermo Fischer FTIR 6700 Fourier
transform infrared spectrometer setup in the attenuated total reection
(ATR) mode with a silicon crystal. The spectra were measured on
three samples on the surface for each data point. The sample FTIR
spectra are shown in Figure S2. The ratio between the absorbance of
the vinyl group (809 cm1) with respect to the carbonyl group (1705
cm1) is used to measure the monomer conversion percentage similar
to the technique used in a reference.40 The reaction conversion is
obtained by using the starting ink (Avinyl/Acarbonyl)ink as 0% conversion
reference and the samples with four curing passes with the UV fusion
(Avinyl/Acarbonyl)UVfusion as the fully cross-linked reference.
H DOI: 10.1021/acsami.7b10443
Research Article
Self-Folding Electronic Composite Characterization and
Video. The supply voltage for the electrochromic composite was
provided using a Sourcemeter 2611B (Keithley Instruments,
Cleveland, OH). Images and videos were recorded with Canon EOS
60D DSLR.
ASSOCIATED CONTENT
*
S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acsami.7b10443.
Swelling of UV-cured polymer samples in dierent
components (Figure S1); sample FTIR spectra (Figure
S2); dierential scanning calorimetry (DSC) and
dynamic mechanical analysis (DMA) data (Figure S3);
mechanical strain tests of the stress layer material (Figure
S4); modeling the folding in the bilayer geometry
(Figure S5); eect of heating and high UV dose on the
folding behavior (Figure S6); design of the electronic
composite (Figure S7); folding angle with stops (Figure
S8); unfolding forces (Figure S9) (PDF)
Video S1. Self-folded electrochromic device (MPG)
Video S2. Unfolding experiments (MPG)
AUTHOR INFORMATION
Corresponding Author
*E-mail: subras@mit.edu.
ORCID
Subramanian Sundaram: 0000-0002-8456-916X
Ryan C. Hayward: 0000-0001-6483-2234
Author Contributions
S.S. conceived the project and was involved in all of the aspects
of the work. S.S. and R.C.H. planned the experiments. D.S.K.
helped with performing some characterization tests. S.S. wrote
the manuscript with input from all of the authors. All of the
authors discussed the results and read the manuscript.
Notes
The authors declare no competing nancial interest.
ACKNOWLEDGMENTS
This work was supported by the DARPA SIMPLEX program
through SPAWAR under contract no. N66001-15-C-4030. The
views expressed are those of the authors and do not reect the
ocial policy or position of the DoD or the U.S. Government.
We thank Nicholas G. Bandiera for careful reading of the
manuscript and discussions.
REFERENCES
(1) Whitesides, G. M.; Grzybowski, B. Self-Assembly at All Scales.
Science 2002, 295, 24182421.
(2) Hawkes, E.; An, B.; Benbernou, N. M.; Tanaka, H.; Kim, S.;
Demaine, E. D.; Rus, D.; Wood, R. J. Programmable Matter by
Folding. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 1244112445.
(3) Felton, S.; Tolley, M.; Demaine, E.; Rus, D.; Wood, R. A Method
for Building Self-folding Machines. Science 2014, 345, 644646.
(4) Guo, X.; Li, H.; Ahn, Y. B.; Duoss, E. B.; Hsia, K. J.; Lewis, J. A.;
Nuzzo, R. G. Two- and Three-Dimensional Folding of Thin Film
Single-Crystalline Silicon for Photovoltaic Power Applications. Proc.
Natl. Acad. Sci. U.S.A. 2009, 106, 2014920154.
(5) Hayes, G. J.; Liu, Y.; Genzer, J.; Lazzi, G.; Dickey, M. D. Self-
Folding Origami Microstrip Antennas. IEEE Trans. Antennas Propag.
2014, 62, 54165419.
(6) Song, Y. M.; Xie, Y.; Malyarchuk, V.; Xiao, J.; Jung, I.; Choi, K.-J.;
Liu, Z.; Park, H.; Lu, C.; Kim, R.-H.; Li, R.; Crozier, K. B.; Huang, Y.;
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX
ACS Applied Materials & Interfaces
Rogers, J. A. Digital Cameras with Designs Inspired by the Arthropod
Eye. Nature 2013, 497, 9599.
(7) Zotov, S. A.; Rivers, M. C.; Trusov, A. A.; Shkel, A. M. Folded
MEMS Pyramid Inertial Measurement Unit. IEEE Sens. J. 2011, 11,
27802789.
(8) Xu, S.; Yan, Z.; Jang, K.-I.; Huang, W.; Fu, H.; Kim, J.; Wei, Z.;
Flavin, M.; McCracken, J.; Wang, R.; Badea, A.; Liu, Y.; Xiao, D.;
Zhou, G.; Lee, J.; Chung, H. U.; Cheng, H.; Ren, W.; Banks, A.; Li, X.;
Paik, U.; Nuzzo, R. G.; Huang, Y.; Zhang, Y.; Rogers, J. A. Assembly of
Micro/Nanomaterials into Complex, Three-Dimensional Architectures
by Compressive Buckling. Science 2015, 347, 154159.
(9) Studart, A. R.; Erb, R. M. Bioinspired Materials that Self-Shape
through Programmed Microstructures. Soft Matter 2014, 10, 1284
1294.
(10) Felton, S. M.; Tolley, M. T.; Shin, B.; Onal, C. D.; Demaine, E.
D.; Rus, D.; Wood, R. J. Self-Folding with Shape Memory Composites.
Soft Matter 2013, 9, 7688.
(11) Zhao, Q.; Zou, W.; Luo, Y.; Xie, T. Shape Memory Polymer
Network with Thermally Distinct Elasticity and Plasticity. Sci. Adv.
2016, 2, No. e1501297.
(12) Overvelde, J. T. B.; de Jong, T. A.; Shevchenko, Y.; Becerra, S.
A.; Whitesides, G. M.; Weaver, J. C.; Hoberman, C.; Bertoldi, K. A
Three-Dimensional Actuated Origami-Inspired Transformable Meta-
material with Multiple Degrees of Freedom. Nat. Commun. 2016, 7,
No. 10929.
(13) Na, J.-H.; Evans, A. A.; Bae, J.; Chiappelli, M. C.; Santangelo, C.
D.; Lang, R. J.; Hull, T. C.; Hayward, R. J. Programming Reversibly
Self-Folding Origami with Micropatterned Photo-Crosslinkable
Polymer Trilayers. Adv. Mater. 2015, 27, 7985.
(14) Mu, J.; Hou, C.; Wang, H.; Li, Y.; Zhang, Q.; Zhu, M. Origami-
Inspired Active Graphene-based Paper for Programmable Instant Self-
Folding Walking Devices. Sci. Adv. 2015, 1, No. e1500533.
(15) Blees, M. K.; Barnard, A. W.; Rose, P. A.; Roberts, S. P.; McGill,
K. L.; Huang, P. Y.; Ruyack, A. R.; Kevek, J. W.; Kobrin, B.; Muller, D.
A.; McEuen, P. L. Graphene Kirigami. Nature 2015, 524, 204207.
(16) Scott, T. F.; Schneider, A. D.; Cook, W. D.; Bowman, C. N.
Photoinduced Plasticity in Cross-Linked Polymers. Science 2005, 308,
16151617.
(17) Liu, Y.; Boyles, J. K.; Genzer, J.; Dickey, M. D. Self-Folding of
Polymer Sheets using Local Light Absorption. Soft Matter 2012, 8,
17641769.
(18) Liu, Y.; Miskiewicz, M.; Escuti, M. J.; Genzer, J.; Dickey, M. D.
Three-Dimensional Folding of Pre-Strained Polymer Sheets via
Absorption of Laser Light. J. Appl. Phys. 2014, 115, No. 204911.
(19) Schmidt, O. G.; Eberl, K. Nanotechnology: Thin Solid Films
Roll Up into Nanotubes. Nature 2001, 410, 168.
(20) Xu, B.; Hayward, R. C. Low-Voltage Switching of Crease
Patterns on Hydrogel Surfaces. Adv. Mater. 2013, 25, 55555559.
(21) Ge, Q.; Qi, H.; Dunn, M. L. Active Materials by Four-
Dimension Printing. Appl. Phys. Lett. 2013, 103, No. 131901.
(22) Raviv, D.; Zhao, W.; McKnelly, C.; Papadopoulou, A.; Kadambi,
A.; Shi, B.; Hirsch, S.; Dikovsky, D.; Zyracki, M.; Olguin, C.; Raskar,
R.; Tibbits, S. Active Printed Materials for Complex Self-Evolving
Deformations. Sci. Rep. 2014, 4, No. 7422.
(23) Tibbits, S. 4D Printing: Multi-Material Shape Change. Archit.
Des. 2014, 84, 116121.
(24) Tumbleston, J. R.; Shirvanyants, D.; Ermoshkin, N.;
Janusziewicz, R.; Johnson, A. R.; Kelly, D.; Chen, K.; Pinschmidt,
R.; Rolland, J. P.; Ermoshkin, A.; Samulski, E. T.; DeSimone, J. M.
Continuous Liquid Interface Production of 3D Objects. Science 2015,
347, 13491352.
(25) Ladd, C.; So, J.-H.; Muth, J.; Dickey, M. D. 3D Printing of Free
Standing Liquid Metal Microstructures. Adv. Mater. 2013, 25, 5081
5085.
(26) Kolesky, D. B.; Truby, R. L.; Gladman, A.; Busbee, T. A.;
Homan, K. A.; Lewis, J. A. 3D Bioprinting of Vascularized,
Heterogeneous Cell-Laden Tissue Constructs. Adv. Mater. 2014, 26,
31243130.
I DOI: 10.1021/acsami.7b10443
Research Article
(27) Eckel, Z. C.; Zhou, C.; Martin, J. H.; Jacobsen, A. J.; Carter, W.
B.; Schaedler, T. A. Additive Manufacturing of Polymer-Derived
Ceramics. Science 2016, 351, 5862.
(28) Kokkinis, D.; Schaffner, M.; Studart, A. R. Multimaterial
Magnetically Assisted 3D Printing of Composite Materials. Nat.
Commun. 2015, 6, No. 8643.
(29) Mao, Y.; Yu, K.; Isakov, M. S.; Wu, J.; Dunn, M. L.; Qi, H. J.
Sequential Self-Folding Structures by 3D Printed Digital Shape
Memory Polymers. Sci. Rep. 2015, 5, No. 13616.
(30) Gladman, A. S.; Matsumoto, E. A.; Nuzzo, R. J.; Mahadevan, L.;
Lewis, J. A. Biomimetic 4D Printing. Nat. Mater. 2016, 15, 413418.
(31) Zhang, Q.; Zhang, K.; Hu, G. Smart Three-Dimensional
Lightweight Structure Triggered From a Thin Composite Sheet via 3D
Printing Technique. Sci. Rep. 2016, 6, No. 22431.
(32) Mao, Y.; Ding, Z.; Yuan, C.; Ai, S.; Isakov, M.; Wu, J.; Wang, T.;
Dunn, M. L.; Qi, H. J. 3D Printed Reversible Shape Changing
Components with Stimuli Responsive Materials. Sci. Rep. 2016, 6,
No. 24761.
(33) Wu, J.; Yuan, C.; Ding, Z.; Isakov, M.; Mao, Y.; Wang, T.; Dunn,
M. L.; Qi, H. J. Multi-Shape Active Composites by 3D Printing of
Digital Shape Memory Polymers. Sci. Rep. 2016, 6, No. 24224.
(34) Ge, Q.; Sakhaei, A. H.; Lee, H.; Dunn, C. K.; Fang, N.; Dunn,
M. L. Multimaterial 4D Printing with Tailorable Shape Memory
Polymers. Sci. Rep. 2016, 6, No. 31110.
(35) Espalin, D.; Muse, D. W.; MacDonald, E.; Wicker, R. B. 3D
Printing Multifunctionality: Structures with Electronics. Int. J. Adv.
Manuf. Technol. 2014, 72, 963978.
(36) Ota, H.; Emaminejad, S.; Gao, Y.; Zhao, A.; Wu, E.; Challa, S.;
Chen, K.; Fahad, H. M.; Jha, A. K.; Kiriya, D.; Gao, W.; Shiraki, H.;
Morioka, K.; Ferguson, A. R.; Healy, K. E.; Davis, R. W.; Javey, A.
Application of 3D Printing for Smart Objects with Embedded
Electronic Sensors and Systems. Adv. Mater. Technol. 2016, 1,
No. 1600013.
(37) MacDonald, E.; Wicker, R. Multiprocess 3D Printing for
Increasing Component Functionality. Science 2016, 353, No. aaf2093.
(38) Zarek, M.; Layani, M.; Cooperstein, I.; Sachyani, E.; Cohn, D.;
Magdassi, S. 3D Printing of Shape Memory Polymers for Flexible
Electronic Devices. Adv. Mater. 2016, 28, 44494454.
(39) Rodriguez, J. N.; Zhu, C.; Duoss, E. B.; Wilson, T. S.;
Spadaccini, C. M.; Lewicki, J. P. Shape-Morphing Composites with
Designed Micro-Architectures. Sci. Rep. 2016, 6, No. 27933.
(40) Huang, L.; Jiang, R.; Wu, J.; Song, J.; Bai, H.; Li, B.; Zhao, Q.;
Xie, T. Ultrafast Digital Printing Toward 4D Shape Changing
Materials. Adv. Mater. 2017, 29, No. 1605390.
(41) Ding, Z.; Yuan, C.; Peng, X.; Wang, T.; Qi, H. J.; Dunn, M. L.
Direct 4D Printing via Active Composite Materials. Sci. Adv. 2017, 3,
No. e1602890.
(42) Sundaram, S.; Jiang, Z.; Sitthi-Amorn, P.; Kim, D. S.; Baldo, M.
A.; Matusik, W. 3D-Printed Autonomous Sensory Composites. Adv.
Mater. Technol. 2017, 2, No. 1600257.
(43) Sitthi-Amorn, P.; Ramos, J. E.; Wangy, Y.; Kwan, J.; Lan, J.;
Wang, W.; Matusik, W. MultiFab: A Machine Vision Assisted Platform
for Multi-Material 3D Printing. ACM Trans. Graphics 2015, 34,
129:1129:11.
(44) Glugla, D. J.; Alim, M. D.; Byars, K. D.; Nair, D. P.; Bowman, C.
N.; Maute, K. K.; McLeod, R. R. Rigid Origami via Optical
Programming and Deferred Self-Folding of a Two-Stage Photo-
polymer. ACS Appl. Mater. Interfaces 2016, 8, 2965829667.
(45) Timoshenko, S. Analysis of Bi-Metal Thermostats. J. Opt. Soc.
Am. 1925, 11, 233.
(46) Walker, S. B.; Lewis, J. A. Ink Composition for Making a
Conductive Silver Structure. US 2015/0004325 A1, 2015.
(47) Andersson, P.; Nilsson, D.; Svensson, P.-O.; Chen, M.;
Malmstrom, A.; Remonen, T.; Kugler, T.; Berggren, M. Active Matrix
Displays Based on All-Organic Electrochemical Smart Pixels Printed
on Paper. Adv. Mater. 2002, 14, 14601464.
ACS Appl. Mater. Interfaces XXXX, XXX, XXXXXX